Introduction

Object recognition is a basic application domain of image processing and computer vision. For many decades it has been – and still is – an area of extensive research. The term “object recognition” is used in many different applications and algorithms. The

  • PDF / 369,068 Bytes
  • 10 Pages / 439.37 x 666.142 pts Page_size
  • 40 Downloads / 188 Views

DOWNLOAD

REPORT


Introduction

Abstract Object recognition is a basic application domain of image processing and computer vision. For many decades it has been – and still is – an area of extensive research. The term “object recognition” is used in many different applications and algorithms. The common proceeding of most of the schemes is that, given some knowledge about the appearance of certain objects, one or more images are examined in order to evaluate which objects are present and where. Apart from that, however, each application has specific requirements and constraints. This fact has led to a rich diversity of algorithms. In order to give an introduction into the topic, several areas of application as well as different types of requirements and constraints are discussed in this chapter prior to the presentation of the methods in the rest of the book. Additionally, some basic concepts of the design of object recognition algorithms are presented. This should facilitate a categorization of the recognition methods according to the principle they follow.

1.1 Overview Recognizing objects in images is one of the main areas of application of image processing and computer vision. While the term “object recognition” is widely used, it is worthwhile to take a closer look what is meant by this term. Essentially, most of the schemes related to object recognition have in common that one or more images are examined in order to evaluate which objects are present and where. To this end they usually have some knowledge about the appearance of the objects to be searched (the model, which has been created in advance). As a special case appearing quite often, the model database contains only one object class and therefore the task is simplified to decide whether an instance of this specific object class is present and, if so, where. On the other hand, each application has its specific characteristics. In order to meet these specific requirements, a rich diversity of algorithms has been proposed over the years. The main purpose of this book is to give an introduction into the area of object recognition. It is addressed to readers who are not experts yet and should help them to get an overview of the topic. I don’t claim to give a systematic coverage M. Treiber, An Introduction to Object Recognition, Advances in Pattern Recognition, C Springer-Verlag London Limited 2010 DOI 10.1007/978-1-84996-235-3_1, 

1

2

1

Introduction

or even less completeness. Instead, a collection of selected algorithms is presented attempting to highlight different aspects of the area, including industrial applications (e.g., measurement of the position of industrial parts at high precision) as well as recent research (e.g., retrieval of similar images from a large image database or the Internet). A special focus lies on presenting the general idea and basic concept of the methods. The writing style intends to facilitate understanding for readers who are new to the field, thus avoiding extensive use of mathematics and compact descriptions. If suitable, a link to som

Data Loading...